Why it matters
An assistant builds its picture of you from many pages at once: your site, directories, press, app stores, an old pricing page you forgot existed. When those disagree, the model picks one version, states an outdated fact, or hedges. A wrong price or a wrong “not available in Colombia” in an answer costs you the buyer before you ever hear about it.
You cannot argue with the model. You can make the truth consistent and easy to find on the pages it reads. Google documents Organization structured data as one way to state name, description and official profiles unambiguously; the rest is keeping your facts in one place and your pages in line with it.
The workflow
Write the brand truth once
What the business does, for whom and where. The current goal. The positioning: how you differ from the alternatives, in one paragraph. Tone and writing rules. Pricing exactly, with currency and the date it applies from. Notes on each competitor. This is the reference every claim is checked against.
Name your key pages and give each a role
The money page, the comparison page, the hub page, and the rest. Every fact should have one page that is its home, so a correction always has somewhere to go.
Audit what answers actually say about you
Ask the tracked buyer questions plus direct ones: “how much does X cost”, “does X work in Colombia”, “who is X for”. Grade each claim about you as correct, incomplete, outdated, conflicting or unverifiable. Compare pricing exactly, not approximately.
Trace each wrong claim to its source
Look at the pages the answer cites: a stale directory profile, a comparison article from two pricing changes ago, a partner page that describes an old plan. A claim that appears in two or more answers, or on two surfaces, is not a fluke and is worth fixing.
Correct the source, not the answer
Update your own page first, with the approved fact in plain words. Ask the third party to update theirs. Add Organization structured data and date your pricing pages so an old copy is easy to recognise as old. A human publishes every correction; there is no shortcut into the model.
Keep Context and FactCheck current in VozRadar
Context holds the brand truth and the key pages; FactCheck grades what answers claim about you and turns repeated wrong claims into opportunities with the evidence and the approved fact attached.
Run it in VozRadar
Context + FactCheck
- Context holds Business, Current goal, Positioning, Tone & writing, competitor notes and key pages with a role; onboarding prefills it and never overwrites your edits
- The Brand truth panel in FactCheck holds approved facts; “What AI answers claim about you” grades each claim correct, incomplete, outdated, conflicting or unverifiable, comparing pricing exactly
- A conflicting or outdated claim seen in at least 2 answers, or on at least 2 surfaces, becomes an opportunity: “Correct a wrong claim in AI answers” or “Correct outdated pricing in AI answers”, with the evidence and the approved fact
- Fixes, fix prompts and the content brief use the context, and the fix prompt includes the approved fact and the key page for that topic; a human publishes the correction
In the app: Project → Context and Project → FactCheck. Correction opportunities appear under Fixes.
The context fields to fill:
- Business: what you sell, to whom, in which markets.
- Current goal: the one outcome this quarter is about.
- Positioning: how you differ from the alternatives, in a paragraph.
- Tone & writing: the rules any generated draft must follow.
- Competitor notes: one line per competitor on where you win and where you do not.
- Key pages: each with a role: Money page, Comparison, Hub or Other.
See how AI sees your brand.
Start with one market. Expand when you're ready.